“How Human Is Human? The View from Robotics Research” by Ishiguro Hiroshi

Ishiguro Hiroshi Ishiguro Hiroshi

This curious little book by Japanese technologist Ishiguro Hiroshi, now available in a very readable English translation by Tony Gonzalez, nominally discusses what robotics research teaches us about what it means to be human. But one can’t help but be left with the impression that what it really shows is just how different Japan can at times be from other parts of the world.

These differences start with technology itself. This is not a book about autonomous behavior, artificial intelligence, speech or vision recognition, deep learning or any of the other subjects that exercise English-language discussions of the human/machine continuum. Instead, the robots under study here are all tethered to an individual human operator in some way, extending, as it were, human reach. Ishiguro’s androids converse, give lectures or act in plays, but only because there is a human somewhere at the controls.

Technically, Ishiguro is right that there is nothing in the term “robot” that requires autonomous action or decision-making, yet “dumb robot” would in West now normally be seen, if not as an oxymoron, then perhaps at least a bit 20th-century. Even a Roomba can figure out on its own how to cover a room and not get stuck in corners.

Ishiguro is more concerned about how people—and operators—interact with androids. This is not without interest: the mechanics (motors, air actuators) needed to get an android to appear humanlike are complicated and, one imagines, marvels of mechanical engineering. Some of the details, like emulating breathing, seem strikingly pointless. However, interestingly if not surprisingly, somewhat humanlike robots are less well-received than ones that don’t look human at all:

 

Even if it has a perfectly humanlike appearance, an android that moves awkwardly looks very creepy, almost like a zombie.

 

Ishiguro also discusses how operators can, after a while, find themselves considering the remote androids almost as extension of their own bodies and selves:

 

operating a Geminoid is very simple; it’s largely a matter of watching some computer monitors and pushing some buttons while you speak with people. After using a Geminoid for a while, you start to feel like it’s your own body, and those you’re interacting with start to consider the Geminoid you.

 

How Human Is Human? The View from Robotics Research, Ishiguro Hiroshi, Tony Gonzalez (trans) (Japan Library, March 2020)
How Human Is Human? The View from Robotics Research, Ishiguro Hiroshi, Tony Gonzalez (trans) (Japan Library, March 2020)

This is, up to here, all par for the course. But Ishiguro then proceeds to makes an android that looks just like himself, which leads him to conclude

 

I sometimes feel that Geminoids are not just similar to their models but are actually more human than we are.

 

In what seems to be a bit like a solution looking for a problem, Ishiguro starts giving lectures remotely using the android, although if the conference is overseas, it is only practical to send the head. Finally, the android starts receiving conference invitations itself, addressed to “Dear Copy of Prof Ishiguro”.

Concerned that his android self is better-looking than he is, Ishiguro then loses weight and undergoes cosmetic surgery:

 

Having regained a younger appearance so that I once again looked like Geminoid H1-1, I had recovered the identity my android had taken from me.

 

But still, aging takes its toll: “If possible,” he writes, “I wish to remain forever identical to Geminoid H1-1.”

 

Even allowing for the possibility that Ishiguro is having us on, the Western reader, at least this one, is left blinking at all this, pulling up Pygmalion, Avatar and Westworld from one’s cultural memory. Once one has gotten over the distinct weirdness of a researcher mechanically cloning himself and writing about it (and worrying about such details as whether hair implants or wigs look better on his duplicate self), one can then engage with some of the underlying assumptions which have evidently sent Japanese researchers in different directions than their non-Japanese counterparts.

Ishiguro and, presumably, large parts of the Japanese tech community, evidently think that physically humanoid devices will play a large role in human-machine interaction. Western research is, for the most part, going in an different direction, focusing on, for example, visual projections and other sorts of digital simulations rather than physical emulations. Japan has long led in mechanical, rather than digital, engineering: it is not surprising, therefore, that their efforts go into motors and silicon skin rather than, say, holographic projection or virtual reality displays.

Is Ishiguro right? In these pandemic-stricken days, androids might allow one work from home while also being in the office. Or one could just use Zoom.


Peter Gordon is editor of the Asian Review of Books.